Right now, Object types are defined as classes containing only the properties metadata. However, after parsing happens, we could think of enriching the object automatically before storage in the database. This would belong to the transformation in an ETL pipeline.
I can image defining a method enrich() that takes care of this:
class Instrument(ObjectType):
# PropertyTypeAssignments here
def enrich(self, ...) -> None:
# some logic here to enrich the object type without needing to type all metadataThis is equivalent to the dynamic properties in openBIS, but written in Python, instead of Jython. Thus it would allow us to use Python libraries to automate it (e.g., ase for unit cell symmetry analysis, or perhaps also weldx?).
This is similar to the normalize() function in NOMAD. The idea is that some properties might be automatically extracted based on some metainformation already available in the object or in parent/child objects, making things more automatically, hence with a better UX.
Right now, Object types are defined as classes containing only the properties metadata. However, after parsing happens, we could think of enriching the object automatically before storage in the database. This would belong to the transformation in an ETL pipeline.
I can image defining a method
enrich()that takes care of this:This is equivalent to the dynamic properties in openBIS, but written in Python, instead of Jython. Thus it would allow us to use Python libraries to automate it (e.g., ase for unit cell symmetry analysis, or perhaps also weldx?).
This is similar to the
normalize()function in NOMAD. The idea is that some properties might be automatically extracted based on some metainformation already available in the object or in parent/child objects, making things more automatically, hence with a better UX.